Tuesday, September 18, 2007

Does Christianity make people soft-hearted? I often wonder how much influence religion really has on people. The General Social Survey asks people how well does the term soft-hearted describe them. Their answers range from "not well" (1) to "very well" (5). Here are the means by religion:

Mean soft-heartedness score

Inter-denominational 4.37
Protestant 4.23
Christian 4.16
Catholic 4.08
Jewish 3.95
None 3.91
Muslim 3.69
Buddhist 3.12

Clearly, Christians are the most compassionate. From knowing people, I would have guessed Jews.


  1. Certainly self reporting can't be the best measure of soft heartedness or compassion. Seems more likely that the list reflects how much value the group's sect places on the public appearance of compassion and soft heartedness.

    "Inter denominationals" are likely to be Unitarians, some form of Universalists or "crypto Calvinists" (ty Mencius Moldbug), secular Jews who attend Quaker schools in hip neighborhoods and other leftist "social justice" types. They are basically defined by how they BELIEVE themselves to be more compassionate than everyone else and love telling everyone else how compassionate they are.

  2. I'm not too surprised. Jews are very intolerant of those they think are anti-semitic. Christians, on the other hand are enjoined to turn the other cheek and love your enemy.

  3. Anonymous10:13 AM

    Subjective answers like these are meaningless. They might display certain patterns, but you can't draw concrete conclusions such as "Since this group of persons say X, then that must mean they are Z."

  4. Anon: You mean that you are unable to tell us validly anything about yourself?

  5. Anonymous3:09 PM

    Ron, your initial question is causal, your data, correlational. Nice leap! As for your response to the other anonymous (not me), there is a difference between self-ratings and behaviors. If I am soft-hearted in my own mind, but a serial killer, which data do you value more? When the data conflict, we value measures of behavior more than subjective measures of self. Ron, your response mischaracterizes anonymous' claim. The data are UNRELIABLE, not necessarily invalid; however, they can be both.

  6. Anon: We do the best we can in the absence of an experiment.

    I said nothing about behavior--I am speaking about one's feelings.

    The first Anon said the measure is meaningless; that refers to validity, not consistency. He didn't say something like you would get a different answer from everyone tomorrow. He is claiming that it does not get at what it attempts.

  7. Anonymous3:38 PM

    Ron, in the absence of evidence, we withhold conclusion, we don't say "oh well, we don't have an experiment" and then draw a causal inference. There ARE experimental data out there, you simply chose the conveniently available non-experimental data. To claim that you have done the best you can, is inaccurate.

    As for the second point, anon1's point is on the spot. Because a person says he is good-hearted does not make him so. Most people believe that they are better than the average in many ways, yet that does not make it so. Have you also heard of the fundamental attribution bias (error)? People judge themselves differently than they judge others. First, we judge others on their behaviors, not their feelings. Second, in Western cultures, when we judge the behaviors of others we tend to attribute their actions to personal characteristics rather than to the situation. When we do good things, we attribute that to who we are. When we do something wrong, we tend to blame the situation. Behaviors matter. Being "soft-hearted" is demonstrated by actions, not by thoughts. Therefore, your question regarding whether you can say anything "validly" (however you meant that) about yourself, the answer is most likely not, but we do the best we can with the available machinery (our brains).

    As for the final point. BOTH invalid and unreliable measures render data, analysis of those data, and conclusions based on the analysis of invalid and/or unreliable data, meaningless. Validity refers to whether a measure actually measures what it purports to measure (and therefore influences the conclusions); reliability refers to how well the measure measures whatever it is measuring (which may or may not be what it purports to measure) and therefore influences the conclusions on draws from the data.

  8. Anon: You are not withholding conclusions in the absence of evidence. I, at least, have data: you, on the other hand, have speculation, and have based your conclusions on that.

    Data trumps speculation, so bring us some of that so-called experimental data that tests the soft-heartedness of people from various religions, and we'll talk.

    Attribution bias refers to how we explain behavior, not how we characterize ourselves and others. We tend to explain why we do good things in terms of our character, and in terms of circumstances when others do something good, and conversely for bad behavior.

    Where is your evidence that Western culture does not recognize feelings, etc.? When I read the survey question, I thought about feelings, and last time I checked I wasn't Chinese.

    If the question elicited only noise, why don't we see equal means, roughly? Why are the random Chistian noises lined up together?

    All of a sudden, people are erecting very high standards for drawing tenative conclusions about things. If people haven't picked it up yet, this blog is (mostly) about data. I generally follow the data, and if someone shows up with better data, I'm all ears.

  9. Anonymous8:14 PM

    Ron, I am not going to get into a pissing match here, but I will address two of your points simply because they interest me. Your comment regarding attribution bias is accurate, however, somewhat curiously you say "Attribution bias refers to how we explain behavior,not how we characterize ourselves and others. We tend to explain why we do good things in terms of our character.." So we explain behavior in terms of our character, but that attribution bias is irrelevant regarding how we characterize ourselves? We characterize ourselves and others in terms of our behaviors all the time. If there is a bias in HOW we attribute behavior, then this will have an influence on how we characterize ourselves.

    As for the core issue that led me to post in the first place, "Does Christianity make people soft-hearted?" you make a causal inference from correlational data. This is a logical fallacy.

    Regarding why the data might line up the way they do... Christians have a pre-occupation with actions (aka behavior). The data you present might simply reflect that Christians BELIEVE that they are doing good and that doing good reflects softheartedness. There are a myriad of explanations that the data support, but a very softened version of yours is no better than the rest (that being that "Christians tend to rate themselves as being more soft-hearted than other religions."). Heck, maybe soft-hearted people become Christians! The data support exactly that claim as well. The two basics about correlation that students learn in a basic statistics/research methods class is that the direction of causation in a correlation typically cannot be determined and even if logic suggested that one preceded the other, you cannot rule out a third (or fourth or fifth) influence. Alas, we come full circle to why your data do not support your claim. I am not saying your claim is wrong, but the data you have don't allow you to make such a claim!


Meta-analysis of clinical trials: Eat walnuts

I am always looking for easy eating choices that are good for you. This new meta-analysis of 26 clinical trials looked to see if walnuts ma...